Modeling Number of Claims and Prediction of Total Claim Amount


Acar A., KARABEY U.

International Conference on Numerical Analysis and Applied Mathematics (ICNAAM), Rhodes, Greece, 19 - 25 September 2016, vol.1863 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 1863
  • Doi Number: 10.1063/1.4992295
  • City: Rhodes
  • Country: Greece

Abstract

In this study we focus on annual number of claims of a private health insurance data set which belongs to a local insurance company in Turkey. In addition to Poisson model and negative binomial model, zero-inflated Poisson model and zero-inflated negative binomial model are used to model the number of claims in order to take into account excess zeros. To investigate the impact of different distributional assumptions for the number of claims on the prediction of total claim amount, predictive performances of candidate models are compared by using root mean square error (RMSE) and mean absolute error (MAE) criteria.